Effect of O2 , H2 +O2 , and N2 +O2 ion-beam

JOURNAL OF APPLIED PHYSICS 109, 114317 (2011)

Effect of O21, H211 O21, and N211 O21 ion-beam irradiation on the field emission properties of carbon nanotubes J. J. S. Acunã,1 M. Escobar,2,3 S. N. Goyanes,3 R. J. Candal,2,4 A. R. Zanatta,5 and F. Alvarez1,a) 1

Instituto de Fı´sica “Gleb Wataghin,” UNICAMP, P.O. Box 6165 Campinas, SP, 13083-970, Brazil INQUIMAE, FCEyN-UBA-CONICET, Ciudad Universitaria, Pabelloń II, Buenos Aires, Argentina 3 Depto. Fı´sica, FECyN, UBA, Ciudad Universitaria, Pabelloń II, Buenos Aires, Argentina 4 ECyT, 3iA, UNSAM, Campus Migueletes, San Martıń, Pcia. Buenos Aires, Argentina 5 Instituto de Fı´sica de Saõ Carlos-USP, P.O. Box 369, Saõ Carlos 13560-250, Brazil 2

(Received 31 January 2011; accepted 23 April 2011; published online 15 June 2011) The effect of O2þ, H2þþ O2þ, and N2þþ O2þ ion-beam irradiation of carbon nanotubes (CNTs) films on the chemical and electronic properties of the material is reported. The CNTs were grown by the chemical vapor deposition technique (CVD) on silicon TiN coated substrates previously decorated with Ni particles. The Ni decoration and TiN coating were successively deposited by ion-beam assisted deposition (IBAD) and afterwards the nanotubes were grown. The whole deposition procedure was performed in situ as well as the study of the effect of ion-beam irradiation on the CNTs by x-ray photoelectron spectroscopy (XPS). Raman scattering, field-effect emission gun scanning electron microscopy (FEG-SEM), and field emission (FE) measurements were performed ex situ. The experimental data show that: (a) the presence of either H2þ or N2þ ions in the irradiation beam determines the oxygen concentration remaining in the samples as well as the studied structural characteristics; (b) due to the experimental conditions used in the study, no morphological changes have been observed after irradiation of the CNTs; (c) the FE experiments indicate that the electron emission from the CNTs follows the Fowler-Nordheim model, and it is dependent on the oxygen concentration remaining in the samples; and (d) in association with FE results, the XPS data suggest that the formation of terminal quinone groups decreases the CNTs C 2011 American Institute of Physics. [doi:10.1063/1.3593269] work function of the material. V

I. INTRODUCTION

At present, it is recognized that carbon-based materials in the form of nano-tubes, -onions, and -domes have an enormous potential for field-emission devices applications.1,2 Carbon nanotubes (CNTs), in particular, are interesting materials with various applications in materials science and electronics.3,4 More recently, the incorporation of substitutional atoms in the graphene network has been proposed as an alternative route to modify its properties. The inclusion of nitrogen, for example, is expected to change the electronic properties of the material by acting as a donor impurity when incorporated into a graphiticlike structure.1,5 However, the effective control over the properties of either N-doped or pure CNT’s is critical and requires removal defective structures typically present in these materials.6 Previous reports confirm that Oþ ion-beam irradiation of nanotubes preferentially eliminates carbon atoms bonded to highly reactive parts of the material by forming volatile carbon oxide compounds.7 Reports relating ion beam purification of nanotubes is not common in the literature.1,7 Studies of graphite ion oxygen beam erosion can help to give some clues about the phenomenon involves in the oxygen nanotubes interaction. However, these studies are generally performed by energetic ions (>1 keV) in the attempt to understand the effect of O2þ a)

Electronic mail: [email protected].

0021-8979/2011/109(11)/114317/7/$30.00

irradiation on graphite in fusion experiments, and consequently, a straight forward comparison with our results is difficult.8 Although this limitation, our findings show that the low energy O2þ bombardment produces volatiles compounds (CO2 and CO) as reported with energetic O2þ ions probably by eliminating defective C sites. The etching process takes place because well-organized structures such as fullerenes, nano-domes, -tubes, -onions, and -horns containing graphitic planes are thermodynamically more stable than amorphous (a-)C or other defective structures.9–11 We note that, in spite of the extensive use of etching purification process, the mechanisms leading to a better material are not fully understood. Consequently, the study of the microscopic mechanisms involved in the oxygen ion-beam irradiation of nanostructured graphene-based materials could improve our understanding as well as lead to further control of their electronic properties.12 The effect of oxygen etching can be seen, for example, in Fig. 1 where a CNT structure presents flaws and dangling bonds. These defects can react with oxygen species, which are either eliminated or, when remaining attached, act as network terminators. Based on the above facts, the aim of this paper is to investigate in situ the effect of O2þ, H2þþ O2þ, and N2þþ O2þ ion-beam irradiation on the electronic properties of CNTs deposited by CVD. The analysis of the physical-chemical properties of the nanostructures produced by the ion-beam

109, 114317-1

C 2011 American Institute of Physics V

Downloaded 15 Jun 2011 to 143.106.72.169. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

114317-2

Acunã et al.

J. Appl. Phys. 109, 114317 (2011)

FIG. 1. (Color online) (a) Schematic CNT’s showing flaws and dangling bonds. (b) After ion-beam irradiation, oxygen species (open circles) are bonded to these defects. (c) Further sample processing (such as thermal annealing, for example), eliminates very defective regions of the CNT and some oxygen acts as bond terminators (open circles).

irradiation was studied by XPS in a ultrahigh vacuum (UHV) chamber directly attached to the deposition and irradiation chamber, i.e., the samples were analyzed by XPS without atmospheric contamination.13 Afterwards, and in order to study the changes in both structural and electronic properties of the material, ex situ Raman scattering, electron microscopy, and field emission experiments were performed. II. EXPERIMENTAL DETAILS

The CNTs are synthesized by CVD using acetylene as a source of carbon on an in situ decorated catalyst Ni nanoparticles (NPs) substrate. In fact, the substrate consisted of mirror polished n-type (15 X/cm2, ) silicon wafers coated with TiN. The purpose of the TiN film (100 nm thick) is to avoid the formation of nickel silicide and it was grown by ion-beam assisted deposition (IBAD).14 Briefly, a high-purity titanium target (99.9995%) was sputtered by argon ions (1450 eV, 13 mA/cm2) in a nitrogen atmosphere (partial pressure: 4.8 102 Pa) at 650 C. Afterwards, the Ni NPs were deposited by sputtering a Ni target (99.9995%) which was followed by annealing the sample for 5 min at 650 C. Further details on the Ni island preparation can be found elsewhere.15 After the catalyst NPs preparation, the temperature of the substrate was raised to 700 C and the CNTs were grown during 5 min by introducing a gaseous mixture of [H2]:[Ar]:[C2H2] ¼ 1:10:6 sccm inside the preparation chamber (kept at 25 Pa). As we shall see below, “carpet like” nanotubes are obtained from this procedure. Three different series of nanotubes samples were then irradiated in situ by O2þ, H2þþ O2þ, N2þþ O2þ ion-beams. The chamber partial pressures due to the O2, H2, and N2 gases feeding the Kaufman guns was maintained at 1.7 103, 1.4 102, and 1.9 102 Pa, respectively. The nominal energy of the ion-beam irradiation was fixed at 30 eV and the current-density beam was 1.4 mA/cm2, i.e., the current measured by the electronic ion gun instrument divided by the area of the Kaufman exit grid of 3 cm diameter (Oxford Instrument). Therefore, this grid determines the density of current at the barley exit of the gun. Considering that in the experiment the samples is located 20 cm from the gun exit grid, the self ion repulsion spread up the beam to a diameter of 10–12 cm lowering the density of current roughly one order of magnitude, i.e., the beam at the sample is 0.14 mA/cm2. The selection of 30 eV ion energy is a compromise between stability of the ion beam and low energy bombardment of the sample to avoid strong sample damage. Finally, each irradiation procedure is followed by annealing the samples at 400 C for 10 min (in order to remove volatile com-

pounds) and afterwards transferred to the XPS analysis chamber where their structural-chemical characteristics are investigated. The XPS spectra were taken using the Al Ka line ( ¼ 1486.6 eV) and a VG-CLAMP-2 electron hemispherical analyzer14 providing a spectral resolution of 0.85 eV. The atomic composition of the samples was determined by integrating the core-level peaks, properly weighted by the photoemission cross section.16 The inelastic scattering background contributing to the peaks associated with the electron corelevel spectra was subtracted by using Shirley’s method.17 Afterwards, the spectra were adjusted by a standard multiples 50%–50% Gaussian-Lorentzian peaks fitting procedure. The ex situ, Raman spectra were obtained in the backscattering geometry by exciting the samples with the 488.0 nm at roomtemperature. Scanning electron microscopy (FEG-SEM GEMINI DSM 982) was employed for further characterization of the nanostructures. The SEM images were essentially performed with the main goal to observe possible changes due to the irradiation treatment. Some TEM images (High-Resolution TEM, JEOL JEM-3010 URP) were obtained in few samples with the purpose to determine if the catalyze was on the top or in the bottom of the nanotube, i.e., to discard that some of the observed effects of the field emission experiments were originated in metallic particles. Indeed, the TEM images show that the nickel particles remain in the bottom of the nanotube, i.e., they continue stacked in the substrate after sample growth. It is important to remark that TEM experiments determining the number of nanotubes before and after irradiation as well as their size and type (single wall, multi wall) could bring interesting physical insight on the material properties. At the present we have not systematic TEM images of the samples and more work is necessary before drawing conclusions regarding with the catalyst mechanism of the process. At last comment regarding the catalyst. In accordance with the TEM results, most of the nickel particles remain in the root of the nanotubes. Therefore, the possible formation of metallic compound with O, N, and H will not affect the field emission results. The field emission characteristics of the samples were studied in a sphere-to-plane electrode configuration, with an anode radii of 1.01 mm, at 22 C, under vacuum conditions (1.3 104 Pa). The current versus voltage curves (I V) are obtained under different electrodes separation and normalized. The I V data are well represented by assuming a Fowler-Nordheim emission mechanism. In the case of a parallel plane geometry, if d represents the distance anode-cathode, the density of current J as a function of the electric field E0 ¼ V=d is given by JðE0 Þ ¼ aE20 expðg=E0 Þ, where, V is


114317-3

Acunã et al.

J. Appl. Phys. 109, 114317 (2011)

the applied voltage, a ¼ ab2 /1 , and g ¼ b/3=2 b1 . Here, b and / are the so-called enhancement geometrical factor and work function of the material, respectively. The constants a 3 and b are usually pffiffiffiffiffiffiffiffi given, in the F-N equation, by a ¼ e =8ph and b ¼ 8p 2me =3eh, where e and me are the charge and electron mass, and h is the Plank’s constant. By direct substitution one can obtain a ¼ 1.54 106 [A eV V2] and b ¼ 6.83 109 [eV3/2 V m1]. In the configuration of sphere-to-plane setup electrodes such as those used in this paper, a correction of the F-N equation is necessary and the following expression is obtained for the I V relationship:18,19 I ¼ 2prV

a V 2 gd : exp g d V

(1)

This equation is valid provided that r d, where r is the radius of the anode tip. The so-called linearized F-N form is written as a function of d=V as: 2 Id a d g ; (2) ¼ ln 2p ln rV 3 g V where g is the slope of the straight line represented by Eq. (2). Defining IðV=dÞ ¼ Aeff J ðV=dÞ, with r and d geometrical known parameters, by fitting the experimental I V data one can obtain the value of g and subsequently the effective emission area given by: Aeff ¼

2prV 2pbrV ¼ : g b/3=2

(3)

As observed in Eq. (3) the effective emission area Aeff depends on the applied voltage due to the fact that the anode is curved.18 It is common to establish the relationship between / and b from the slope of Eq. (2) or the interception of the curve to the ordinate axis. However, previous reports suggest that the former procedure does not always give reliable results.20,21

FIG. 3. (Color online) Raman scattering spectra of CNTs irradiated in situ with O2þ, H2þþ O2þ, and N2þþ O2þ ions. D and G stand for the disorder- and graphitelike vibrational modes, respectively.

Therefore, in this paper we used the ordinate at the origin to calculate the relationship / ¼ /ðbÞ, where / is given by: /¼

2pab3 bexpðfÞ

2=5 ;

(4)

where f ¼ lnð2pa=gÞ is the intercept defined by Eq. (2). III. RESULTS A. Morphological and structural characteristics of the CNTs

Figure 2 presents FEG-SEM micrographs of all the studied samples. Figure 2(a) corresponds to the pristine sample. From this picture (and the deposition time), the growth rate was estimated to be 280 nm/min. The picture shows the carpetlike CNTs in the back plane. The white parts observed in Figure 2(a) correspond to CNTs that were peeled off and moved from a different part of the sample. Figures 2(b), 2(c) and 2(d) correspond to samples irradiated with O2þ, H2þþ O2þ, and N2þþ O2þ ion-beam, respectively. These pictures show that the main features of the original samples are preserved in spite of the irradiation procedure. Figure 3 shows the Raman scattering spectra corresponding to samples irradiated with different ion-beam conditions. The spectra show the so-called D-band (1365 cm1) and G-band (1590 cm1) that are associated with the presence of structural disorder (mode A1g) and graphite (mode E2g), respectively.22,23 The Raman spectra present features that depend on the ion-beam characteristics, which are related to the oxygen incorporation.

TABLE I. Atomic concentration (as obtained from XPS) of the studied CNT samples: before and after irradiation. Sample

FIG. 2. FEG-SEM images of CNT samples: (a) pristine, and (b) irradiated with (c) O2þ, H2þþ O2þ, and (d) N2þþ O2þ.

Pristine O2þ irradiated H2þþ O2þ irradiated N2þþ O2þ irradiated

Carbon (at.%)

Oxygen (at.%)

100 81 95 39

0 19 5 61


114317-4

Acunã et al.

J. Appl. Phys. 109, 114317 (2011) TABLE II. Summary of the different contributions present in the XPS spectra deconvolution.

FIG. 4. (Color online) Representation of ln(Id2/rV3) versus d/V curves, for all studied CNT samples, before and after irradiation.

B. Compositional and structural studies

The carbon and oxygen atomic concentration as obtained from the XPS analysis is shown in Table I. First of all, it must be said that, in the case of the N2þþ O2þ ion-beam the nitrogen concentration remains below 0.5 at.%, i.e., the limit of our detection technique. Also, we have noticed that the presence of H2þ in the ion-beam diminishes the amount of oxygen in the material. On the contrary, N2þ ions enhanced the oxygen incorporation, which is even higher than that provided by irradiating the samples with pure O2þ ions. Considering that the XPS technique probes only the most external surface of the samples (up to 5 nm) one can assume that, most probably, these results correspond to the tips of the nanotubes. C. Field emission results

Figure 4 presents the ln(Id2/rV3) versus d/V curves for all of the studied samples. As described in Sec. II this behavior is characteristic of the Flower-Nordheim model, i.e., the transport mechanism is a tunneling emission phenomenon.24 According to the literature, it is common to define the emission threshold field as the applied (macroscopic) electric field (E0 ) necessary to produce a fixed current density.

FIG. 5. ID/IG peak ratio obtained from the D- and G-bands (Fig. 3) after a two Gaussian fitting procedure of the studied samples.

Band

Binding energy (eV)

Chemical bonding

PC1 PC2 PC3 PC4 PO1 PO2 PO3

284.4 6 0.1 285.3 6 0.1 287.5 6 0.1 290.1 6 0.1 530.3 6 0.2 531.7 6 0.2 533.2 6 0.2

aromatic CAC [1] CAC sp3 type [1,2,13] C @ O, CAO, and COO [1,2] p-plasmons and/or shake-up [15] >C @ O quinone [38] C @ O in esters, amides and anhydrides [31] AOH [1]

However, this definition is not adequate considering that the cathode geometry can influence the results.25 As a result, in order to characterize the behavior of the electronic emission, we propose that it is more reliable to calculate the current density JL obtained when an arbitrary defined local electric field, namely, EL ¼ 1 V/nm, is applied. Here EL ¼ bE0 , and b, and E0 were previously defined. Experimental evidence show that EL is typically a few V/nm, and it is significantly higher than E0 .25,26 Table III shows all the parameters defined in Sec. II and obtained from fitting the Eq. (2) to the experimental results displayed in Fig. 4. IV. DISCUSSION A. Morphological and material structural characteristics

As observed from the Raman scattering spectra (Fig. 3), the ion-bean irradiation treatments reduce the contribution of the D-band while the G-band is not modified. In other words, the disorder-associated vibrational modes are the ones mostly affected by the treatment. The effect of the ion-beam characteristic on the decreasing of the ID/IG ratio (Fig. 5) suggests an efficient elimination of the more reactive parts of the material—essentially dangling and wrong bonds. Moreover, the Raman spectra present some broadening of the high-frequency side of the G- and D-bands, most probably due to the incorporation of oxygen in the CNTs (Table I).

FIG. 6. (Color online) XPS spectrum and PCi fitting components associated with the C1s core level. The spectrum corresponds to a CNT sample irradiated with N2þþ O2þ ions.


114317-5

Acunã et al.

J. Appl. Phys. 109, 114317 (2011)

FIG. 7. (Color online) XPS spectra associated to the emission of O1s electrons in a CNT sample after irradiation with O2þ, H2þþ O2þ, and N2þþ O2þ ions. The circles correspond to the experimental data, and the straight lines to fitting functions (associated to C ¼ O ketone or quinine, C ¼ O carbonyl, and -OH groups).

B. Effect of beam irradiation on the material chemical structure

As mentioned in Sec. III B, the fitting procedure of the structure associated with the band due to C1s electrons yields four bands: PCi with i ¼ 1, 2, 3, 4 (Table II). All the studied samples present C1s XPS spectra similar to the one shown in Fig. 6, which corresponds to the CNT sample irradiated with N2þþ O2þ ions. Here, PC1 (284.4 eV) is attributed to CAC bonds in graphitic sp2 configuration; PC2 (285.3 eV) is related to sp3 bonded carbon atoms; PC3 (287.5 eV) is due to carbon in carbonyl (C @ O